Mortgage churning preventable with right AI

  • By Kate Weber

Finding overlooked customers and engaging with them is vital to decrease mortgage churning amongst the industry. 

CEO of artificial intelligence SaaS Elula, Josh Shipman said currently 700,000 home loan accounts churn each year in Australia. 

Indeed, it was an issue that big lenders grappled with the recent RFi Group Mortgage Innovation Summit.

Shipman said “explainable AI” can be used to drive better consumer outcomes especially surrounding customer retention within a bank's home loan portfolio. 

Predicting why a consumer will leave is just as important as understanding who will leave added Shipman.   

“It's about unpacking the black box of machine learning.” 

Pinpoint accuracy can be found in Elula’s machine learning as it can find targeted conversations or a targeted customer list rank from most ‘sticky’ to ‘least sticky’.

From here AI brings analytics to the table that focuses on the customers a company need to speak to with tailored conversations.

“The conversation with customers, and getting that right is absolutely critical.

“There's a lot of customers that often get overlooked, which are those customers that are much more loyal." 

Shipman said translating data and information into everyday context so the end user in the contact centre can have a business related as to why that customer will leave. 

“We want to be able to take the higher volume, low cost interactions with those customers and reward that loyalty as well.” 

Shipman added having the right conversation at that critical time in the customer's life is going to help drive a lot better value back. 

Organisations of any size and budget should have access to meaningful ROI to deliver results back to them said Shipman, yet much of the machine learning struggles are some of the hardest to solve. 

“It's a little bit like turning up to divorce proceedings with a bunch of flowers, you've missed that opportunity.” 

“You're really searching for a needle in a haystack. 

“It’s on many levels with lots of complexity. Businesses take lots of different approaches to retention. “

Shipman said one of the first issues companies face is being caught off guard when a customer is going to leave but fall short as current methods may need to be updated.  

Trying to reclaim a client in this phase of their journey is pointless as many have already made up their minds. 

“You can be the best salesperson in the world and talking them off that ledge is really difficult. 

“It's a little bit like turning up to divorce proceedings with a bunch of flowers, you've missed that opportunity.” 

Awakening “sleeping dogs” is another dangerous attempt at keeping customer satisfaction as it can inspire people to look at better rates elsewhere. 

“If you look at all of your customers who've actually churn it's less than 20 per cent of those customers, which means almost 80 per cent of those phone calls shouldn't have been made in the first place.” 

Traditional analytics is another area that fails the mortgage broker industry as a standard propensity to churn model or other factors can lead to limited data set that are unable to be updated and maintained accordingly. 

Data privacy also plays a large part in the Elula process given the senitive nature of the information provided said Shipman. 

"Privacy is it's such an important topic, as is security.  

"Within the Elula environment everything that is sent to us is tokenised.

“We are unable to identify who that customer is. So that is contractually set by us at the beginning. So sort of ground rules that says we can't know customers coming in, but we provide a key and we can do the translation on your site to then de-identify customers.

"That's a really critical point. So we do no joining of our data," concluded Shipman.